Land cover classification and ecosystem monitoring with satellite observation
Land cover classification by means of Earth observation constitutes one of the most well-established operational applications of satellite remote sensing. Multispectral optical sensors and synthetic aperture radar (SAR) systems allow land use and land cover categories to be discriminated at periodicities compatible with the phenological cycles of vegetation. The combination of both modalities is particularly relevant in environments with high cloud cover, where the penetration capability of radar complements the limitations of the optical spectrum. Time series of images, processed by means of supervised and unsupervised classification algorithms, enable the generation of dynamic cartographies that document the evolution of vegetation cover, agricultural surfaces, urban areas and wetlands at regular intervals. The monitoring of wetland ecosystems presents specific technical requirements arising from the seasonal variability of the water surface and the density of emergent vegetation. Spectral indices derived from the near-infrared and short-wave infrared, together with SAR interferometric coherence, allow the flooded extent, surface biomass and hydrological recharge cycles to be characterised. This information is operationally relevant for water resource management, the assessment of the conservation status of protected habitats and compliance with reporting obligations under European environmental directives such as the Water Framework Directive. The detection of changes in long time series introduces radiometric and geometric consistency considerations that condition the comparability between acquisitions from different platforms or periods. Surface reflectance normalisation, atmospheric correction and precise scene-to-scene registration are critical pre-processing stages that determine the statistical reliability of change indicators. In contexts of support for public policy, the methodological traceability and the uncertainty associated with classifications must be rigorously documented in order to underpin administrative decisions or reporting commitments to international bodies. From the perspective of critical infrastructure operators and public administrations, the integration of land cover products into territorial information systems allows risks associated with ecosystem degradation, the loss of productive land or the alteration of hydrological regimes to be anticipated. The availability of open-access satellite data, combined with processing capabilities on cloud computing platforms, has reduced the operational barriers to the systematic generation of these products. Nevertheless, the technical interpretation of results and their translation into management indicators require specialised competencies that remain the differentiating factor in the Earth observation value chain.
NASSAT - Network Satellite Systems